Sample Roasting with Kaffelogic Nano

tl;dr: Crop to Cup has added a powerful new sample roaster to our line up and have developed profiles to support our lab which can be found HERE

When Crop to Cup’s sister company, Showroom Coffee, brought the Kaffelogic Nano 7 to the U.S. market, our staff had the opportunity to play, tinker, and learn about the small but incredibly capable countertop roaster from New Zealand. This new and affordable air roaster presents a unique opportunity for us to raise the bar on quality both internally and externally. For us, one of the most exciting prospects is that the Kaffelogic, which promises exceptional and repeatable roasts at a fraction of the price of other roasters, could make sample roasting accessible and more affordable to farmers and other producing partners, further democratizing evaluation of quality.

Additionally, putting the Kaffelogic through its paces over thousands of cycles would give us a good opportunity to reflect on how we approach sample roasting in the Crop to Cup lab.

With this in mind, we set out to develop a Kaffelogic profile for 50 gram roasts utilizing the optional boost kit which provides the ability to have full control over load size from 50-200 grams. While larger batch sizes might appeal more to home roasters, in a lab setting, 50 grams offers the advantage of preserving as much sample weight as possible. Preservation of sample material cushions any roast issues we may encounter, saves green for evaluation against new offers, pre-shipment samples, and arrival comparisons, and gives us more sample weight to socialize with our team and customers alike. Smaller batch sizes also save time as a smaller batch cools faster—a welcome improvement during busy offer seasons where we receive hundreds of samples each week.

We looked for the roast profile to hit a target relative development desired by our lab for each sample we roast; the ideal roast profile would give us “the best look” at each coffee. While this can be very subjective, our perception of a “best look” of a sample reveals the complexity and intensity of its unique flavor notes, acidity, body, sweetness, and aftertaste. A non-exhaustive list of measurements we take to compare relative developments of roasts are: weight loss, total roast time, end temperature, and development time.

All the tools we use to evaluate these measurements can be found in our Lab Supplies.

Our next requirement was that the profile must suit a next-day lab cupping setting in order to improve our speed of evaluation, supplier feedback and purchasing decisions. For this particular profile project, the roast quality we are looking for does not depend on resting time or brewing outside of a cupping environment.

Lastly, the profile should be easily mutable and able to accommodate for coffees of varying density, processing, bean size, variety, moisture content, and water activity. We did not expect that one single profile will be able to accommodate for each of these variables, and in fact, we ended up developing three different profiles for different ranges of bulk fill density.

One of the most, if not the most crucial characteristics for this project was that the roast profile should be consistent and repeatable, and produce roasts that tasted consistent with each other.

Our initial plan was as follows:

• Trial a series of profiles using the same coffee, keeping the boost kit set to 50g and roast number to 1.0.
• Analyze roast data collected and compare roasts using cupping lab sensorial analysis.
• Select the best-tasting profiles and deconstruct the roast settings to develop our own lab profile that best suits our needs.

How that adventure worked out in practice:
We began by selecting a test coffee. We decided to use a washed Ethiopian coffee for testing as we have done extensive roast trials of this specific coffee on other machines and knew what it is capable of tasting like. Further, washed Ethiopian coffees have fairly consistent and predictable behavior in the roaster, which would make dialing in the profile much easier. This particular coffee also has a clear acidity and fruit quality that is known to us, and a roast that is “in focus” would be more readily apparent to our cupping team.

It is worth noting this project was tested on cutting-edge software and firmware with the roaster units running the latest (as of writing) firmware, 7.13.15 and Kaffelogic Studio 6.3.9.3. By chance, this roast project happened to fortuitously coincide with the roll out of Kaffelogic’s new live logging feature which allows the optional monitoring of roasts as they happen on a computer, which gave great real time feedback and greatly expedited profile development.

We trialed the standard profiles found on the machine as well as selected a few profiles to test from the Kaffelogic community at large, including Christopher Feran’s Kaffelogic sample profile. The community profiles we looked into all revolved around a core “descending ROR” roasting style, which we find preferable and more reliable for sample roasting. The incredibly lively and responsive KL forum community was a core part of this initial profile selection process. The deconstruction, analysis, and in-depth discussion of these profiles give us a strong starting point, and were very informative. In my opinion, the online community that surrounds the KL is one of this roaster’s greatest strengths. We compared these selected profiles to a few core profiles that we had experience with, with the expectation that all of the profiles will behave differently with the 50g boost kit setting.

A few profiles failed initially due to observable roast outcomes that were not to our standard (under/over-developed, tipped, etc.) and adjustments had to be made to the profiles to bring them closer into the specific development window we were aiming for. From this initial cupping we selected the top 3 profiles to continue making adjustments, to begin to develop our profile.

The first major adjustment we decided to make was to shorten the profiles to a roast time that was more appropriate for our smaller batch size. Through the next series of cuppings, we found that our sweet spot for the 50g batch size was consistently a little over 6 minutes. With our target roast time set, we now could do some roast curve sculpting. Our next series of cuppings focused on tasting variations in development time and weight loss of different roast curves within that target ~6min roast time. We found ~50 seconds of development time and 11-12% weight loss preferable for our 50g batch size. While we were getting closer to our desired profile, the primary issues we faced started coming into focus.

First, our small batch size made our roasts much more sensitive to excessive fan speed. We observed that a few roast tests would come out extremely light while still having weight loss in excess of 13%. Our conclusion was that the fan speed was too high, and we made adjustments accordingly, which solved this issue. Second, we were beginning to perceive roast differences between roasts with the exact same profile and green coffee. In the drum roaster world, consistency issues are usually resolved by developing a BBP (between batch protocol), which helps you control and keep consistent the thermal momentum between roasts. The Kaffelogic cool-down procedure and “roast-from-cold” capabilities limit our options for BBP, so we had to look towards other parameters we could adjust to control our energy between roasts. At this stage meeting virtually with the KL team was pivotal, and provided a lot of guidance. We ran an experiment testing variations on the preheat temperature parameters and found that a higher preheat temp gave us more consistent roasts, without experiencing any overheating or unevenness in internal bean development. With these adjustments, our roast profile started to come into its own. With some final tweaks to our roast end temperature we had our profile dialed in for our trial coffee.

The next step was to translate this profile to and prove it with other coffees.

We tested ~15 different coffees from our sample library of varying origins, moisture contents, processing, and bulk fill densities (as measured using a Lighttells MD500). We found that the greatest determining factor in comparative roast quality using the same profile was the bean density, and to a lesser extent, processing style. We expected this; fluid bed roasters’ use of convective heat transfer as the primary roasting mechanism interacts with coffee depending on its volume in the roaster, morphology and rate of absorption. Bean density not only will determine how much fan power we need to agitate the bean mass, but also how aggressive our roast curve ought to be to fully and evenly develop the beans. Low density coffees contain less mass and thus receive a higher-per-gram amount of energy using the same profile relative to higher density coffees and thus will lose their water faster and become brittle quicker and more evenly than higher density beans. Low density coffees will also be more easily agitated by the same amount of airflow (fan speed) than high density coffees would. Increased agitation will make the coffee harder to develop on the same profile, and the combination of these two opposing phenomena results in the strange results we saw earlier (high weight loss, while still being relatively light). Processing also plays a minor role in this, as natural and experimental fermentation processes develop relatively faster and more evenly than washed coffees, and we should expect to accommodate for this.

Since our test coffee was a washed, high density Ethiopian coffee, translating this profile to other coffee types involved reducing our fan speed and making our roast curve more aggressive, which would be more appropriate for lower density beans. Dialing these in took several trials (including blind cuppings with several trained cuppers both involved with and not involved with the project), but we eventually were able to develop three total profiles based on 3 density ranges using readings from our MD500; <790g/L, 790-820g/L, and >820g/L. We will likely continue to experiment with these three base profiles as we encounter new and exciting coffees that lay outside of our test range.

We encourage users to adjust the roast level to their preference and liking (our recommended setting is at 2.0 to allow for greater development flexibility), experiment, share their results with the Kaffelogic community at large. Any feedback is welcome and we are, as always, excited to continue to learn and grow with this amazing community.

Happy roasting!

– Erinn Buhyoff